Re: [syzbot] [mm?] WARNING in mas_nomem

On 26/07/31 12:30AM, syzbot wrote:
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    cc2b5f627e8c Add linux-next specific files for 20260714
> git tree:       linux-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=15ce5632580000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=2290ccbf984c524f
> dashboard link: https://syzkaller.appspot.com/bug?extid=78a7af91b68fc6394771
> compiler:       Debian clang version 22.1.8 (++20260613092233+e80beda6e255-1~exp1~20260613092250.77), Debian LLD 22.1.8
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=179151b9580000
> 
> Downloadable assets:
> disk image: https://storage.googleapis.com/syzbot-assets/6836f8efb1da/disk-cc2b5f62.raw.xz
> vmlinux: https://storage.googleapis.com/syzbot-assets/0109d3477cc7/vmlinux-cc2b5f62.xz
> kernel image: https://storage.googleapis.com/syzbot-assets/c2e49e350bbf/bzImage-cc2b5f62.xz
> 
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: [email protected]
> 
> ------------[ cut here ]------------
> !mas->sheaf && !mas->alloc
> WARNING: lib/maple_tree.c:5751 at mas_nomem+0x1cc/0x210 lib/maple_tree.c:5751, CPU#0: udevd/5627
> Modules linked in:
> CPU: 0 UID: 0 PID: 5627 Comm: udevd Not tainted syzkaller #0 PREEMPT_{RT,(full)} 
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/16/2026
> RIP: 0010:mas_nomem+0x1cc/0x210 lib/maple_tree.c:5751
> Code: 4c 89 ff e8 d6 85 ea f6 49 c7 07 00 00 00 00 b0 01 e9 95 fe ff ff e8 f3 2d 7e f6 90 0f 0b 90 e9 2b ff ff ff e8 e5 2d 7e f6 90 <0f> 0b 90 e9 77 fe ff ff 89 d1 80 e1 07 80 c1 03 38 c1 0f 8c ce fe
> RSP: 0018:ffffc9000562f8f0 EFLAGS: 00010293
> RAX: ffffffff8b4775cb RBX: 1ffff92000ac5fbf RCX: ffff8880366e3e80
> RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
> RBP: 0000000000000cc0 R08: 00000000ffffffff R09: 1ffffffff1ca1a86
> R10: dffffc0000000000 R11: fffffbfff1ca1a87 R12: dffffc0000000000
> R13: ffffc9000562fe18 R14: ffffc9000562fde0 R15: ffffc9000562fdf8
> FS:  00007f9021203880(0000) GS:ffff8881259f2000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 000056467bb0d750 CR3: 0000000046416000 CR4: 00000000003526f0
> Call Trace:
>  <TASK>
>  mas_store_gfp+0x37a/0x820 lib/maple_tree.c:4987
>  vma_iter_clear_gfp include/linux/mm.h:1674 [inline]
>  do_vmi_align_munmap+0x30c/0x4d0 mm/vma.c:1644
>  do_vmi_munmap+0x252/0x2d0 mm/vma.c:1697
>  __vm_munmap+0x241/0x3e0 mm/vma.c:3347
>  __do_sys_munmap mm/mmap.c:1079 [inline]
>  __se_sys_munmap mm/mmap.c:1076 [inline]
>  __x64_sys_munmap+0x60/0x70 mm/mmap.c:1076
>  do_syscall_x64 arch/x86/entry/syscall_64.c:61 [inline]
>  do_syscall_64+0x17b/0x530 arch/x86/entry/syscall_64.c:85
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
> RIP: 0033:0x7f9021368097
> Code: 73 01 c3 48 8b 0d 61 2d 0d 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 b8 0b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 31 2d 0d 00 f7 d8 64 89 01 48
> RSP: 002b:00007ffeaf32d0a8 EFLAGS: 00000206 ORIG_RAX: 000000000000000b
> RAX: ffffffffffffffda RBX: 000056467bad9bf0 RCX: 00007f9021368097
> RDX: 0000000000000000 RSI: 00000000009480e8 RDI: 00007f90208ba000
> RBP: 000056467bad9910 R08: 000056467baec140 R09: 0000000000000006
> R10: 0000000000000000 R11: 0000000000000206 R12: 00005646569e2588
> R13: 00007ffeaf32d130 R14: 0000000000000000 R15: 0000000000000000
>  </TASK>


Okay, so I'm going to change this warn_on again.

The plan is to:
1. Make mas_erase() and mtree_erase() use __NO_FAIL on retry.
2. Remove the warning so that any failures won't cause the warning.
3. Make mas_store() use __NO_FAIL on retry, like mas_erase().
4. Add the check for locks that cannot sleep to mas_store().

Basically any caller that cannot correctly return the failure will avoid
failing by allocating with __NO_FAIL.  This means the warning is not
necessary.

Thanks to Vlastimil and Pedro for the help on the plan.

Summary of previous conversation is here:
https://lore.kernel.org/all/[email protected]/


Cheers,
Liam